The goal of the ROAM workshop is to introduce our new large-scale benchmark driving dataset, acquired from three diverse cities in South America (currently Peru), which provides a wide range of dangerous driving situations and O.O.D objects that are not found in existing datasets.

Out-of-distribution (OOD) object detection & classification is a vital component to ensure reliability in safety-critical applications such as autonomous driving or medical robotics. Due to its central role, object detection & classification continues to receive significant interest from both academia and industry, particularly by releasing large-scale autonomous driving datasets.  The vast majority of these datasets, however, are typically collected from North American or European cities, while data from the rest of the world is underrepresented. This workshops’ focus is on Robust, Out-of-Distribution, and Multi-Modal models for Autonomous Driving (ROAM) and aims to gather researchers and engineers from academia and industry to discuss the latest advances in OOD perception for autonomous driving.

Please see the About page for a more detailed description of the motivation of the workshop.